Finally, some Australian data on podcasts - but beware of what counts as a 'listener'
When it first arrived, the Podcast Ranker was little help for advertisers. Now it offers download numbers, it has the beginnings of the newest industry trading currency
Welcome to Unmade, mostly written on a chilly Tuesday evening at Sisters Beach, Tasmania, and polished off on a frosty Wednesday morning. There were more whales off the beach yesterday, apparently. Truth be told, I stood on the deck with my binoculars for a few minutes without spotting anything, then gave up and went back to my desk.
Happy National Cheese Toast Day. This feels like a celebration I can participate in.
And there are now 1,122 of us on the Unmade subscriber list. A shoutout to the PHD crew, who signed up en masse yesterday. Funnily enough, PHD was the first media agency to get behind Mumbrella in big numbers back in the day. I always said they were early adopters…
Please do encourage a colleague who’s less au fait with the new than you to get on board.
It’s been a busy week in the audio world. On Monday we saw changes to the radio ratings, which I wrote about in the previous edition of Unmade.
And yesterday, we finally began to see (some) meaningful Australian audience podcast data.
When the Australian Podcast Ranker first launched in September 2019, it had so little information, it did not deserve a great deal of attention. It was literally just a ranking of which podcasts (among those that shared their data) were most downloaded.
But because there were no actual numbers about the downloads or listeners, it seemed to me virtually meaningless. I understood the reason to hold it back - presumably many had such low listenership it might actually put off advertisers.
I’m bloody minded enough that the fact that the stakeholders, the commercial radio industry, had the streaming data, but was not sharing it, made me inclined to ignore the ranker. It wasn’t a very transparent way to begin.
So although it was vaguely interesting - and not surprising to see Hamish Blake and Andy Lee at number one in that first set of data - it did not tell us a lot about the commercial viability of one of my favourite mediums.
The Australian Podcast Ranker also triggered an annoying trend for listeners . If you’ve wondered why every podcast feed suddenly started shovelling out multiple episodes per day, with three minute “bonus” segments or two minute “best bits”, it was because the algorithm rewarded multiple posting into the same feed as it increased download numbers.
Two years on though, as podcasting has continued to grow, provider Triton Digital has now been permitted to share the data. It will provide metrics based on “monthly listeners” and “monthly downloads”. The downloads number seems credible. “Listeners” I’m still not so sure about.
Triton provides its data in accordance with technical guidelines published by the IAB Tech Lab in the US. The document is worth a read, if only to understand just how much guess work is involved. The Apple iPhone is still the dominant podcast listening device, and Apple doesn’t share much data.
So for the most part, the IAB guidelines are a complicated way of saying the the listener number is a guess. It lists the sort of steps that can be taken to prevent the number being ridiculously high - for instance, downloads don’t count if less than a minute’s worth of content was served.
But in the end, there is no way of knowing whether somebody actually listened to a podcast after downloading it. But in the ranker, they will count as a listener either way.
To make matters even more complicated, podcast users can choose whether to automatically download their favourite shows to listen when they’re offline. But they may never get around to it.
Personally, I listen to a number of podcasts every day, but I reckon I hear less than half of my downloaded content. I choose on a whim what to listen to - Wil Anderson and Charlie Clausen’s TOFOP if I’m struggling to sleep (that’s not an insult by the way, I get too annoyed trying to fall asleep to something boring); the ABC’s Coronacast during big days of the pandemic; finance podcast Fear & Greed to start the working day. But not everything that ends up on my phone gets a listen.
And that’s before the fact that the 30-second skip option on the iPhone podcast app makes it easy for listeners to avoid the ads.
Of course, the same issue exists with just about any media measurement. Nobody knows how many people really watched The Masked Singer last night. But the OzTam panel is an attempt to make the best possible guess.
But because the Australian Podcast Tracker is not panel based, it will inevitably over-report listeners. I’m not aware of there being any reliable data being available on the ratio of downloaded minutes of content to a device to the actual number of listening minutes. But if it exists, that could theoretically be applied to understand how many listeners are really out there.
That will be one reason so many commercial messages include a custom URL with the name of the podcast at the end, so sponsors can at least track how much of a response their message drove.
Rather than “listener”, I’d suggest a more accurate label would be intended-to-listen. Logically, a listener would not bother to download the podcast unless they planned to hear it. We just don’t know whether they get round to it.
But with all that said, the ranker does indicate that podcasting is a growth medium, not just because of changing consumer behaviour, but also because the radio companies - who love a growth story for their investors - are leaning into it.
During August, Triton recorded 49m downloads across the publishers participating in the ranking. Which is another point - the ranker is not exhaustive - podcasters are only covered by the data if they choose to be.
For a country with a population of 25m, that number is still relatively low. An average of two downloads per person per month. And no guarantee they listened. Compared to the daily habits of listening to radio, for instance, it suggests podcasting still remains a niche medium.
But of course the rise of smart speakers and the rapid trend towards on-demand content more widely, including video, suggests there’s a lot more growth to come.
So with the caveat that “monthly listeners” aren’t necessarily real listeners, let’s take a look at the data. You can see the full top 100 via this link, by the way.
The weekly Hamish & Andy podcast is still on top. They were the only podcasters to deliver more than 2m monthly downloads in August, which the ranker extrapolates to 919,057 monthly listeners.
But you don’t have to go any further down to find non-radio people. The second most downloaded podcast is the anonymously presented Casefile True Crime, with 1.9m downloads and 860,350.
Third comes Life Uncut, from The Bachelor alumni Brittany Hockley and Laura Byrne. (Yes, I did just have to Google them to discover who they are.) They delivered 1.4m downloads.
Nova’s The Update was fourth, with just under 1.1m downloads, and just under half a million listeners.
The metric now ranks podcasts by listeners rather than downloads, so although Schwartz Media’s 7am podcast did 1.2m downloads, slightly more than The Update, it had fewer listeners, at 352,000.
And only one other podcast had more than a million downloads in the month - The Kyle & Jackie O Show which did 1.6m downloads. This high number was helped by the fact that it pumped out 121 seperate piece of content.
After that, it rapidly becomes a long tail. The numbers - with only six podcasts delivering more than a million downloads per month - suggest it’s a medium that is not yet easy to commercialise.
Those in a niche may find it easier to monetise. To use the Fear & Greed example, their 240,000 monthly downloads -which places them 24th - is a credible enough number for finance marketers looking to tap into a money-focused audience. But for a more generalist audience, radio still offers a hell of a lot more reach.
The data is imperfect, and the word “listener” is contestable. But at least we now have a better picture of the market.
Letters: Listening, not hearing
Yesterday’s edition of Unmade, in which I examined the evolution of the radio ratings system and the weaknesses of the diary system, generated another helpful contribution from my favourite researcher John Grono:
Some thought-starters regarding radio measurement.
FIrst, there is a subtle but an important difference between 'hearing' and 'listening'. 'Hearing' is a limbic function. 'Listening' is a cognitive function.
Take for example when you wend your way through a large shopping mall with all sorts of sounds, music, announcements, Spotify and radio playing. Were you really listening to that 10 seconds of un-recognised music or did you just hear it?
So this is where recall comes into play, which is very important with diaries.
The respondent makes the decision around the distinction between whether they heard a radio station or whether they listened to the radio station. From an advertiser (and hence agency) perspective they want data that most closely correlates to people listening to their ads rather than having just heard their ads.
Of course you can't fill in a diary every minute, hence the 15 minute blocks in which you decide whether you listened to that channel for at least eight of the 15 minutes. It's probably a fair proxy for that person.
Conversely you can get respondents who simply tick the box for every quarter hour of their favourite Announcer/DJ's shift. Some may even do that for every day. So yes, diaries can over-state listening and can (for some) be more like a popularity poll.
Flipping the coin yet again, most people wouldn't carry a diary around with them (umm .. maybe everyone). So at the end of the day, the following morning, or maybe even the end of the week, they fill in the diary. That is really more a test of memory than actual listening and it tends to miss valid listening occasions.
Enter the e-diary on your mobile 'phone. It's a much more immediate and convenient method of data capture and hence likely to be more accurate. Well it is amongst those who are more comfortable with phone apps. But consider how many more elderly people would cope with that method, and would rather a paper diary.
Therefore the 'hybrid' method that the radio industry has been using was a good meld which can be kept representative by varying the proportion of 'phone v. diary as usage patterns change.
Now bring on electronic capture! While there are various ways in which they function they capture sound very accurately. Some systems are more accurate than others (e.g. using a reference system or an embedded code). The AU system is definitely accurate.
But there is one potential issue with electronic capture.
No, it's not the device, the credit rules (e.g. volume), the software etc.
The problems start to happen when we give the devices to people. We tend to forget to wear or carry them from time-to-time. We also forget to charge them.
I had the opportunity to trial one of the devices some time ago. I carried the device and kept a very detailed log for a week down to the minute. Or so I thought.
Yep, the device captured genuine listening that I had missed - mainly radio listening in the car. It also captured listening that I had no cognition of - mainly short snippets of by-passing sound. And, yes, when I went to a pub in Leichhardt to see a band I forgot to take the bloody device.
So if you rely 100% on passive data capture any valid listening when not carrying the device, that data is lost. Forever. With a diary the respondent tends to 'back-fill' based on their memory, or 'usual listening behaviour'.
So I would urge that any respondents who are 100% passive device based have access to a way to 'top-up' their listening manually online.
John Grono, Owner, GAP Research
And Unmade’s update on subscriber numbers triggered a thought too:
When reading the 11th September stories I noticed that Unmade has now 1066 readers registered.
This number is very significant when quoted as a year in English history as the year that the Battle of Hastings occurred. Although that led to many changes - land ownership, inheritance laws, wooden castles to stone - one major change that is still effective today is the language.
William changed the influence from the Germanic inflections more to French. Suffice to say that Unmade is written in the now English language rather than in German.
Just some trivia. The number 1066 just jumped off the page as a long time ago qualified printer.
I welcome your letters to firstname.lastname@example.org, or you can comment via this button
A day of watching video streams lies head. This morning, I’ll be watching a session from The Walkley Foundation on philanthropic models of funding local journalism. It’s a topic I’ll return to in future days.
And this afternoon, we get into TV upfronts season, with Nine first cab off the rank. My gift viewing pack made it all the way to Tasmania. I’ve peeked inside. The box contains Four Pillars gin, which I’ve no complaints about. However, I may not crack it open until after I’ve finished writing up my analysis.
Enjoy your Tuesday.
Proprietor - Unmade